Research Portal

What A Good Wave Is

A wave worth driving for can be small and still have a clean, curling face and a shoulder that peels. Favorable swell direction, light or suitable offshore wind, the seabed and the tide all help determine whether that happens.

Size comes last. A well-shaped two-to-three-foot day can be the best session of the week, and a big, crossed-up, wind-blown one is a wasted drive. That ordering is the goal the surf reading is being built toward: shape with favorable wind and swell direction first, useful period next, size last.

Period describes the time between wave crests passing a point. The buoy number shown here is the dominant period: the period carrying the most wave energy, not the average of every wave. Longer-period waves interact with the seabed in deeper water and can carry more wave power at the same height. Whether that produces a good breaking wave still depends on the beach. Period is not a prediction of set spacing or quiet lulls.

The limit follows from the feeds. No public source Tarline reads reports curl, shape or glassiness; those are read off a wave, not off a model. Direction, period and wind are readable, and they are the conditions that make shape possible, which is as close as the present data gets.

What Runs Today

Current conditions prioritize fresh multi-buoy spectral estimates. Native frequency bins retain their directional moments; an inferred directional distribution is integrated against the beach's exposure and, where a verified coastal origin exists, USGS known-land screening for short-period energy. Alternative buoys contribute a weighted mean rather than stacked copies of the same sea. Arriving energy flux is combined before one Komar and Gaughan (1972) breaker transform.

When that current estimate is unavailable, stale or directionally incomplete, the existing CDIP MOP and exposure-window models remain alternatives. MOP contour-depth heights are de-shoaled before the breaker calculation; the window model applies the beach's directional exposure to offshore forecast components. Future rows use forecast inputs, not extrapolated present buoy observations. Each height keeps the period from the same selected model sample.

The fallback correction system uses beach reports, not agreement with another model. A typed face-height estimate is used directly; a selected height band contributes its approximate midpoint. Those are different levels of precision, and we still need independent observations to evaluate the result. Model-to-model differences are kept separately in Model Diagnostics.

Nearshore and offshore forecast retrieval batches retain their fetch time, so a forecast available before an event can be distinguished from a later revision. The nearshore nowcast table is different: it keeps the latest value for each valid hour, which can be revised. Raw feed snapshots support auditing. The available records can be downloaded from Environmental Data.

The quality reading is a shared 0–100 index of wind, period and usable size. It penalizes surf below 2 ft, more strongly below 1 ft, reduces credit as winds strengthen, and progresses through 8, 12, 16 and 18-second periods. Five dots summarize it in half-dot steps. The map's single dot is red through 35%, orange above 35% through 65%, yellow above 65% but below 75%, and green at 75% and above. Missing required inputs remain Unrated. These are editorial surfing preferences, not observed wave shape, a probability of good surf, or a safety or water-quality assessment.

The Santa Barbara Problem

Santa Barbara makes the local problem especially clear. The Channel Islands and Point Conception shelter much of the coast. Which beaches sit in shadow depends on swell direction and period, and published research shows that sheltered wave energy can be sensitive to small changes in the offshore direction.

Blocking also depends on period, which is why one exposure window per beach is the wrong shape for the problem and why the windows here are banded by period. The windows themselves are editorial priors drawn from that literature and local knowledge. They are starting assumptions to evaluate against observations, not measured beach parameters or proof of forecast accuracy.

What A Skill Claim Would Need

A useful accuracy result needs the forecast available before the event, independent beach observations, whole swell events held out of fitting, and comparison with a named baseline. We need coverage across directions, periods and tides; repeated readings from one event are not independent trials. The required sample size depends on that coverage and the uncertainty in the observations, not a universal count.

Reported surf heights are the input that unblocks it, which is why the report form asks for one. The record so far is public either way: agreement and observation counts in Model Diagnostics, the underlying model and buoy history in Environmental Data, and what each kind of claim requires in Source Investigation.

What Comes Next

The spectral estimate and quality index above are running. The following work targets their remaining uncertainties rather than promising that more data alone will establish accuracy.

1. Matched directional history. The bulk instrument archive is already available. Extend matched native spectral records and audit their gaps, QC flags and deployment changes before evaluating whole swell events against independent observations.
2. Source-audited shallow bathymetry. USGS CoNED provides a concrete high-resolution starting surface. Inspect actual coverage, source dates, void fill and vertical datums before modeling refraction, dissipation or local wave generation. A 1 m grid is not proof of 1 m accuracy.
3. Independent quality evaluation. Compare the published index with dated beach observations of wind, shape and usable waves, keeping typical faces distinct from largest sets. Buoy agreement alone cannot validate those surfing characteristics.

Additional wave simulations, AI models and camera analysis remain possible experiments, not deployed features.